THE ROLE OF PHARMACOGNOSY IN TRADITIONAL AND MODERN SYSTEM OF MEDICINE.pptx
Meaning-in-Language-an-Introduc (2).pdf
1.
2. Series editors
Keith Brown, Eve V.Clark, Jim Miller, Lesley Milroy,
Geoffrey K. Pullum, and Peter Roach
Meaning in Language
An Introductionto Semantics and Pragmatics
3. Meaning in Language
An Introduction to
Semantics and
Pragmatics
D.Alan Cruse
University of Manchester
OXFORD
UNIVERSITY PRESS
8. Contents
Part 2 Words and their Meanings
5 Introduction to lexical semantics
6 Contextual variability of word meaning
7 Word meanings and concepts
8 Paradigmatic sense relations of inclusion and identity
9 Paradigmatic relations of exclusion and opposition
10 Word fields
11 Extensions of meaning
12 Syntagmatic relations
13 Lexical decomposition
Part 3 Semantics and Grammar
14 Grammatical semantics
Parti Fundamental Notions
1 Introduction
2 Logical matters
3 Types and dimensions of meaning
4 Compositionality
Typographic conventions
Preface
ix
x
1
3
17
41
65
83
85
103
125
143
163
177
197
217
237
263
265
9. viii Contents
Part 4: Pragmatics
15 Referenceand deixis
16 Speech acts
17 Implicatures
Conclusion
Answers to questions
References
Index
301
303
329
347
379
383
401
409
10. Typographic conventions
Small capitals
For concepts; occasionally for lexical roots.
Small capitals in square brackets
For semantic components.
Angled brackets
For selectional restrictions
Bold type
For technical termswhen first introduced.
Italics
For citation forms whennot set on a different line.
Bold italics
For emphasis.
Single quotation marks
For quotations from other authors; 'scare quotes'.
Double quotation marks
For meanings.
Question marks
For semantic oddness.
Asterisks
For ungrammaticalityor extremesemantic abnormality.
11. Preface
The aim of this book is not to present a unified theory of meaning inlanguage
(I am not even sure that that would be a worthwhileproject), but to survey the
full range of semantic phenomena, in all their richness and variety, in such a
way that the reader will feel, on completing the book, that he or she has made
face-to-face contact with the undeniably messy 'real world' of meaning. At the
same time, it aims to show that even the messy bits can, at least to some
extent, be tamed by the application of disciplined thinking. As far as seman-
tic theories are concerned, I have been unashamedly eclectic, adopting what-
ever approach to a particular problem seems genuinely to shed light on it.
If there is a theoretical bias, it is in favour of approaches which, like the
cognitive linguistic approach, embrace the continuity and non-finiteness of
meaning.
This is not intended to be a 'baptismal' text; it would probably not be suit-
able for absolute beginners. The sort of readership I had in mind is second- or
third-year undergraduates and beginning postgraduates who have completed
at least an introductory course in linguistics, and who require an overviewof
meaning in language, either as preparation for a more detailed study of some
particular area, or as background for other studies. I would hope it would be
found useful, not only by students of linguistics, but also students of ancient
and modern languages,translation, psychology,perhaps even literature.
Most of the material in the book has grown out of courses in general seman-
tics, lexical semantics, and pragmatics, given to second- and third-year under-
graduates and postgraduates at Manchester University over a number of
years. I owe a debt to generations of students in more than one way: their
undisguised puzzlement at some of my explanations of certain topics led to
greater clarity and better exemplification; critical questions and comments
not infrequently exposed weaknesses in the underlying arguments; and very
occasionally, a genuineflashof insight emerged during a classroom discussion.
The final form of the text was significantly influenced by constructive com-
ments on a draft by Jim Miller of the Universityof Edinburgh, an anonymous
American reviewer,and John Davey of Oxford University Press, although, of
course, full responsibility for remaining imperfectionslieswith myself.
12. Preface xi
The organization of the book is as follows. It is in four parts. Part I dis-
cusses a range of basic notions that underlie virtually all discussions of mean-
ing within linguistics; Part 2 concentrates on aspects of the meanings of
words; Part 3 deals with semantic aspects of grammar; Part 4 introduces the
core areas of pragmatics, and highlights the relations between meaning and
context.
Within Part I, Chapter I provides a very general introduction to questions
of meaning, locating the linguistic study of meaning within the wider context
of the study of signs and communication in general. Chapter 2 introduces a
set of fundamental conceptual tools, mostly drawn from the field of logic,
which, because of their wide currency in discussions of semantic matters,
constitute indispensable background knowledge for a study of meaning in
language. In Chapter 3, a number of concepts are introduced for the descrip-
tion of meanings and differences of meaning. A basic dichotomy (based on
Lyons 1977) is introduced between descriptive and non-descriptive meaning
and, under each of these headings, important types and dimensions of vari-
ation are described. It is rare to encounter any extended treatment of these
topics in semantics textbooks, yet a mastery of them is essential to anyonewho
wishes to talk in a disciplined way about meanings. Chapter 4 discusses the
way(s) in which simpler meanings are combined to form more complex
meanings.
In Part 2, Chapter 5 provides a general introduction to the study of word
meanings, first discussing whether there are any restrictions on what sort of
meanings words can bear, then distinguishing the meaning of a word from that
of a sentence or discourse, and the meanings of full lexical items from the
meanings of grammatical elements. In this chapter the major approaches to
lexical semantics are also outlined. In Chapter 6, the focus is on the range of
variation observable in a single word form in different contexts, ranging from
arbitrarily juxtaposed homonymies to subtle modulations of sense. Chapter 7
introduces a conceptual approach to lexical semantics, beginning with a dis-
cussion of whether and to what extent word meanings can be equated with
concepts. The discussion continues with an outline of prototype theory, the
currently dominant approach to natural conceptual categories, and its rele-
vance for the study of word meanings. Chapters 8 and 9 deal with relations of
sense between lexical items which can occupy the same syntacticposition—in
other words, paradigmatic sense relations, such as hyponymy, meronymy,
incompatibility, synonymy,antonymy, complementarity, reversivity, and con-
verseness. Chapter 10 looks at larger groupings of words—word fields—
mainly structured by the sense relations examined in the previous two chap-
ters. Chapter 11 describes the main types of process, such as metaphor and
metonymy, which enable new meanings to be produced from old ones. In
Chapter 12,meaning relations between words in the same syntactic construc-
tion, that is, syntagmatic sense relations, are examined. Topics discussed
include the nature of normal and abnormal collocations, reasons for a
13. xii Preface
tendency for certain types of words to co-occur, and the nature and con-
sequences of selectional pressures of words on their partners in a string.
Chapter 13 outlines the componential approach to the description of word
meaning, which specifiesmeaning in terms of semantic primitives.
The focus in Chapter 14, which constitutes the whole of Part 3, is on the
sorts of meanings associated with various grammatical entities. First there is a
discussion of the problem of whether there are any constant meanings
attached to categories such as noun, verb, and adjective, and functions such as
subject and object. There then follows a survey of the sorts of meaning borne
by grammatical elements of various sorts, such as number and gender in the
noun phrase, tense, aspect, and modality in connection with the verb, degree in
the adjective, and so on.
Part 4 covers topics which are usually considered to fall under pragmatics,
in that either they involve aspects of meaning which cannot be satisfactorily
treated unless context is taken into account, or they are not propositional in
nature (or both). Chapter 15is concerned with reference, that is,establishing
connections between utterances and the extralinguistic world. Reference is
portrayed as the assigning of values to variables, the variables being signalled
by definite expressions and the values being items in the extralinguisticworld.
Various strategies for indicating (on the part of the speaker) and determining
(on the part of the hearer) correct referents are discussed, including the use
and interpretation of deictic elements, names, and descriptions. Chapter 16
provides an outline of speech act theory, mainly following Austin and Searle
(1969). It discusses the acts that people perform when they are speaking—acts
such as stating, requesting, warning, congratulating, commanding, and so on.
The range of different types of speech act is surveyed and their nature exam-
ined. Chapter 17deals with conversational implicatures, that is, those aspects
of the intended meaningof an utterance which are not encoded in its linguistic
structure, but are, as it were, 'read between the lines'. Different types of con-
versational implicature are described and some proposed explanations of how
they arise are considered.
The concluding chapter briefly surveys the areas covered in the book, sug-
gests practical applications of the study of meaning, and highlights areas
which are currently poorly understood, and where further research is needed.
Each chapter except Chapter I and Chapter 5 contains a set of discussion
questions and/or exercises, suggested answers to which will be found at the end
of the book.
14. Part 1
In this first part of the book, a number of fundamental, but fairly generalnotions are
introduced, which need to be grasped before the more detailed discussions in later
sections can be properly appreciated. Chapter 1 has a scene-setting function, identi-
fying the place of linguistic signs and linguistic communication in the broader
domains of semiotics and communication in general. Chapter 2 introduces a num-
ber of vital conceptual tools drawn from the field of logic. Chapter 3 surveys the
range of different sorts of meaning, and dimensions of variation in meaning. Chapter
4 discusses the notion of compositionality, one of the essential properties of lan-
guage, andits limits.
Fundamental Notions
16. CHAPTER 1
Introduction
1.1 Communication
1.2 Semiotics: somebasic notions
1.3 Language and other communicative
channels
1.4 Characteristics of linguistic channels
5 1.5 Approachesto the study of meaning
7 1.6 The linguisticstudyof meaning
1.7 Branches of the study of meaning
8
Suggestions forfurther reading
9
10
12
15
16
18. CHAPTER 1
Introduction
1.1 Communication
Meaning makes little sense except in the context of communication: the
notion of communication therefore provides as good a place as any to start an
exploration of meaning. Communication can be conceived very broadly,
including within its scope such matters as the transfer of information between
biological generations via the genetic code, the interaction between a driver
and his car, and indeed any sort of stimulus-response situation. Here weshall
confine ourselves to what is surely the paradigm communicative scenario,
namely, the transfer of information between human beings.
1.1.1 Asimple model
Let us begin with a simple model, as shown in Fig. I.I (after Lyons 1977).
In the model, the process begins with a speaker who has something to
communicate, that is, the message. Since messages in their initial form cannot
be transmitted directly (at least not reliably), they must be converted into a
form that can be transmitted, namely, a signal. In ordinary conversation, this
involves a process of linguisticencoding, that is. translating the message into a
linguistic form, and translating the linguisticform into a set of instructions to
the speech organs, which, when executed, result in an acoustic signal. The
initial form of this signal may be termed the transmitted signal.
19. 6 Meaning in language
Every mode of communication has a channel, through which the signal
travels: for speech, we have the auditory channel, for normal writingand sign
language, the visual channel, for Braille, the tactile channel, and so on. As the
signal travels from sender to receiver, it alters in various ways, through distor-
tion, interference from irrelevantstimuli or loss through fading.These changes
are referred to collectively as noise. As a result, the signal picked up by the
receiver (the received signal) is never precisely the same as the transmitted
signal. If every detail of the transmitted signal was crucial for the message
being transmitted, communication would be a chancy business. However, effi-
cient communicating systems like language compensate for this loss of infor-
mation by building a degree of redundancy into the signal. Essentially this
means that the information in a signal is given more than once, or is at least
partially predictable from other parts of the signal, so that the entire message
can be reconstructed even if there is significant loss. It is said that language is
roughly 50per cent redundant.
Once the signal has been received by the receiver, it has to be decoded in
order to retrieve the original message. In the ideal case, the message
reconstructed by the receiver would be identical to the message that the
speaker started out with. Almost certainly, this rarely, if ever, happens; how-
ever, we may presume that in the majority of cases it is 'close enough'. All the
same, it isworth distinguishingthree aspects of meaning:
(i) speaker's meaning:
(ii) hearer's meaning:
(iii) sign meaning:
In the case of an established signalling system like language, the meanings of
the signs are not under the control of the users; the signs are the property of
the speech community and have fixed meanings. Of course on any particular
occasion, the signs used may be ad hoc or conventional, if ad hoc, they may be
prearranged or spontaneous.
speaker's intended message
hearer's inferredmessage
this can be taken to be the sum of the properties
of the signal which make it (a) more apt than
other signals for conveying speaker's intended
message, and (b) more apt for conveying some
messages than others.
1.1.2 Language as a sign system
Any natural human language is a complex sign system, 'designed' to ensure
infinite expressive capacity, that is to say, there is nothing that is thinkable
which cannot in principle be encoded (provided no limit is placed on the
complexity of utterances). Each elementary sign is a stable symbolic associ-
ation between a meaning and a form (phonetic or graphic); elementary signs
may combine together in a rule-governed way to form complex signs which
convey correspondingly complex meanings.
20. Introduction 7
1.2 Semiotics: some basic notions
1.2.1 Iconicity
Signs can generally be classified as iconic or arbitrary. Iconic signs are those
whose forms mirror their meanings in some respect; signs with no natural
analogical correspondences between their forms and their meanings are called
arbitrary. A simple example is provided by the Arabic and Roman numerals
for "three": 3 and III. The Arabic form gives no clue to its meaning; the
Roman version, on the other hand, incorporates "threeness" into its shape,
and is thus iconic. Iconicity is a matter of degree, and usually coexists with
some degree of arbitrariness. Three horizontal lines would bejust as iconic as
the Roman III: the fact that in the Roman symbol the lines are vertical is
arbitrary, as is the fact that its sizecorresponds to that of letters.
Iconicity enters language in several guises. The majority of words in a
natural language are arbitrary: the form of the word dog, for instance, does
not mirror its meaning in any respect. However, the so-called onomatopoeic
words display a degree of iconicity, in that their sounds are suggestive (to
varying degrees) of their meanings:
bang clank tinkle miaow splash cuckoo peewit curlew
whoosh thud crack ring wheeze howl rumble,etc.
The predominance of arbitrariness in the vocabulary is not an accidental
feature, but is a crucial 'design feature' of natural language. There is a limited
stock of dimensions of formal variation in linguistic signs; if all signs were
iconic, it is difficult to seehow universalexpressivitycould be achieved.
Some iconicity is also apparent in grammar. For instance, words which
belong together tend to occur together. In The tall boy kissed theyoung girl we
know that tall modifies boy and not girl because tall and boy come next to each
other in the sentence. In some languages this relationship might be shown by
grammatical agreement, which is a kind of resemblance, and therefore also
iconic. Another way in which iconicity appears in the grammar is that
grammatical complexity by and large mirrors semanticcomplexity.
1.2.2 Conventionality
Many of the signs used by humans in communication are natural in the sense
that they are part of geneticallyinherited biological make-upand do not have
to be learned, although a maturational period may be necessary before they
appear in an individual, and they may be moulded in various ways to fit
particular cultural styles. The sort of signs which are natural in this sense will
presumably include facial expressions like smiling, frowning, indications of
fear and surprise, and so on, perhaps manyof the postural and proxemicsigns
21. 8 Meaning in language
that constitute the so-called 'body language', certain types of gesture, vocal
indications of excitement, desire, etc. (whether or not linguistic), and many
more. Natural signs are likelyto be the most cross-culturally interpretable.
Other signs have conventionally assigned meanings; they have to be specif-
ically learned, and are likely to differ in different communities. Linguistic signs
are the prototypical conventional signs. Even onomatopoeic words usually
have a significant conventional element; often the iconic nature of the word
can only be appreciated, as it were, with hindsight. Take the Turkish word
bulbul. What does it refer to? A baby's babbling? The noise of a mountain
spring? In fact, it means "nightingale". Looking back, one can make the
connection. It is not only linguistic signs that are conventional. Obscene or
offensive gestures, for instance, can vary quite remarkably cross culturally: I
was once reprimanded for pointing the soles of my feet at the Prime Minister
of Iraq (in Arab culture this is disrespectful: my disrespect was entirely
inadvertent). Even in Europe, conventional gestures can differ: Greeks are
famously—and slightly inaccurately—said to shake their heads to say "Yes",
and nod to say "No".
1.2.3 Discreteness
Some signs can vary gradually in their form, and their meanings vary in paral-
lel with the change of form, like the fisherman's indication of the size of 'the
one that got away'; these are called continuous signs. Other signs have fixed
shapes, and must be chosen from a limited inventory: intermediate forms are
not allowed, the range of possibilities is 'chunked'; such signs are described as
discrete. Linguistic signs are virtually all of the discrete variety. Again, this
is not an accidental feature, but has a close connection with iconicity and
arbitrariness: continuous signs are necessarily iconic; arbitrary signs are
necessarily discrete.
1.3 Language and other communicative channels
The prototypical scenario for linguistic communication is two people
engaged in face-to-face conversation. Of course, in such an encounter, lan-
guage signals are exchanged; but normally so are many other types of signal,
and these modify and/or supplement the linguistically encoded message. Let
us, then, briefly look at the semiotic environment of language in a typical
conversation.
The signs that accompany language can be divided into two major types—
paralinguistic and non-linguistic. The defining characteristic of paralinguistic
signs will be taken here to be an extreme dependence on the accompanying
language. Either they cannot be produced except during speech (because they
are carried on the voice), or they cannot be interpreted except in conjunction
22. Introduction 9
with accompanying language. Examples of the first variety are abnormal
volume, tempo, pitch, and voice quality; to function as signs, there must be a
departure from some (personal) baseline or norm. For instance, abnormally
high volume, fast tempo, or high pitch typically signal a heightened emo-
tional state. Examples of the second variety include pausing, emphatic ges-
tures, and gestures which metaphorically depict, for instance, direction of
motion.
The functions of paralinguistic signs can be conveniently classified under
three headings:
(i) Punctuation: there are signs which have functions parallel to those of
punctuation in written language, mainly to segment the stream of
speech so as to facilitate processing.
(ii) Modulation: this involves the addition of an emotive or attitudinal
colouring to the linguistically encoded message.
(iii) Illustration:some signs 'depict' a key element in the message, such as
a direction of movement, or a shape; the depiction may be relatively
literal, like the hand movements of someone describing the climbingof
a spiral staircase, or metaphorical, as when vertical and parallel hands
accompany the setting of limitsof some kind.
Not all the signs that occur alongside language are paralinguistic in the
sense defined. For instance, one may smile or frown while speaking, and this
may well 'modulate' the message. But smiles and frowns (and many other
signs) are perfectly interpretable and capable of being produced in the absence
of any accompanying language. These are therefore to be considered as
non-linguistic.
1.4 Characteristics of linguistic signs
Paralinguistic signs are typically natural, continuous, and iconic, whereas
linguistic signs are for the most part arbitrary, discrete, and conventional.
1.4.1 Simple and complex signs
Linguistic signs may be simple or complex. This does not just mean that they
can occur singly or in groups of various sizes: the occurrence of two or more
signs together does not necessarily result in a complex sign. Take the case of
someone who answers a question with the word Yes,at a higher than usual
pitch, and at the same time smiling. This person has not produced a complex
sign with three constituents, only three simple signs simultaneously. The mean-
ings of the three signs are simply added together: there is no interaction
between the signs other than additivity. Contrast this with a minimally com-
plex sign such as red wine: to obtain the meaning of this sign, we do not simply
23. 10 Meaning in language
add the meaning of red to the meaning of wine (that would give us something
like "wine plus redness"). What happens is the meaning of red interacts with
the meaning of wine by restricting it.
There is no theoretical upper limit to the complexity of linguistic signs. This
is rendered possible by the recursive nature of syntax, that is, the existence of
rules which can be applied indefinitely many times (like the one which yields
This is the dog that worried the cat that killed the rat that ate the com that . . .).
Such rules are an essential prerequisite for the 'universal expressivity' of
language—the fact that anything thinkable is expressible, or at least can be
approximated to any given degree of accuracy.
1.4.2 Signs at different linguistic levels
A linguistic sign may be no more than a phoneme (or two): this is one inter-
pretation of the sl- of slimy, slovenly, slug, slag, slum, slink, slattern, slut, slob,
etc. which seems to indicate something unpleasant, or the gl- of glare, glimmer,
glitter, glisten, glow, gleam, etc. which all have something to do with light
effects. These have no grammatical status, and no contrastive value, but the
intuitions of native speakers leave no doubt that they should receive some
recognition. Other signs occur at higher levels of linguistic organization, from
morpheme level (e.g. the -s of dogs), through word level (e.g. denationaliza-
tion), clause level (e.g. the formal difference between John is here and Is John
here? which signals that one is a question and the other a statement), sentence
level (e.g. We'll do it as soon as you arrive as opposed to As soon as you arrive,
we'll do it), up to text level (e.g. the fact that a stretch of text constitutes a
sonnet is indicated by the form of the text as a whole: this form therefore
constitutes a high-levelsign).
The fact that a sign manifests itself at a particular level does not entail that
it is to be interpreted at that level. A fewexampleswill illustrate this point. The
item the, a word, exerts its semantic effect on a whole noun phrase the littleold
lady who lives in the cottage on the hill; the -ed of John kissed Mary, a bound
morpheme, semantically situates the time relative to the moment of utterance
of the whole event symbolized by John kiss Mary; a single word like matri-
mony maymark a whole discourse as being in a certain register.
1.5 Approaches to the study of meaning
Meaning may be studied as a part of various academic disciplines. There is of
course a significant degree of overlap between disciplines, but characteristic-
ally all have something idiosyncratic and unique in their approach (the
following remarks are merely illustrative).
24. Introduction 11
1.5.1 Philosophy
Linguists typically take the existence of meaning for granted and accept it as
an intuitively accessible 'natural kind'. They do not ask questions like How isit
possible for anything to mean something? or What sort of relation must hold
between X and Yfor it to be the case that X means Y? Such questions are the
province of the philosopher, particularly the philosopher of language.
1.5.2 Psychology
Meaning is a major concern of the psychology of language and psycho-
linguistics. (I shall not attempt to distinguish these.) A distinctive feature here
is the experimental study of how meanings are represented in the mind, and
what mechanisms are involved in encoding and decoding messages. An
example of a fact that could only emerge within a psycholinguistic framework
is that in the lexical decision task, where experimental subjects observe strings
of letters flashed on a screen and must indicate by pressing the appropriate
button whether the string represents a word or not, responses are faster to
words with concrete meanings than to words with abstract meanings, even
when extraneous factors like length and frequency are rigorously controlled.
This observation presumably provides a clue to the role of meaning in word
recognition (to the best of myknowledgeit is still a puzzle).
1.5.3 Neurology
Psychologists take a 'macro' view of mental states and processes. Neurologists,
on the other hand, want to know how these states and processes are imple-
mented at the neuronal level. A psychologist might be broadly compared with
a computer programmer, and a neurologist to the designer of computer chips.
Meaning, like everything else in mental life (at least if one is a physicalist)
must boil down ultimatelyto connections between neurons.
1.5.4 Semiotics
Semioticians view language as one sign system amongst many, and seek out
those features which render it so successful. They are also likely to give
emphasis to marginal aspects of linguistic signification. The recent strong
interest in iconicity in language represents a significant overlap between the
linguistic and semiotic approaches to meaning.
1.5.5 Linguistics
It is not easy to encapsulate the linguistic approach to meaning in language
succinctly. There are perhaps three key aspects. The first is that native
speakers' semantic intuitions are centre-stage, in all their subtlety and
nuances: they constitute the main source of primary data. The second is the
25. 12 Meaning in language
importance of relating meaning to the manifold surface forms of language.
The third isthe respect paid not just to language, but to languages.
1.6 Thelinguistic study of meaningin language
1.6.1 What is linguisticmeaning?
Here we attempt to say what is to count as meaning in language. Following an
impulse towards generosity rather than austerity, we shall as a first step say
that all meaning is potentially reflected in fitness for communicative intent. It
will be assumed that a way of tapping into this is in terms of contextual
normality: every difference of meaning between two expressions will show up
as a difference of normality in some context. Thus, we know that illness and
disease do not mean the same, because during his illness is normal, but during
his disease is not; almost and nearly do not have precisely the same meaning
because very nearly is normal but very almost is not; pass away and kick the
bucket have different meanings because It is with great sadness that we report
that our Beloved Ruler kicked the bucket two minutes after midnightis odd, but
It is with great sadness that we report that our Beloved Rulerpassed away two
minutes after midnight is normal. We take normalityloddness and relative
oddnesslnormality to be primitive intuitions.
It will be noticed that the move in the above characterization was from
meaning to contextual abnormality. Unfortunately, the move cannot without
further ado be made from abnormality to meaning, because there are other
factors besides meaning which affect normality.
Let us assume that we are dealing with spoken language and that the utter-
ance is correctly pronounced. The two sources of abnormality that wewish to
eliminate if possible are grammatical deviance and 'meaning' that is non-
linguistic in origin. Let us make the simplifying assumption that if a pin-
pointed deviance is grammatical in nature, it will not prove possible to reduce
it by contextual manipulation, for instance by interpreting it as metaphor,
science fiction, or fairy-tale. Thus, They goes is irredeemably deviant in any
context, whereas / shallgo there yesterday might just make sense in a setting
where time travel (or at least temporal scrambling) is possible. That leaves non-
linguistic meaning to be taken care of. Consider the possibility that a certain
type of delivery, not amounting to mispronunciation, may be a sign that the
speaker is under the influence of some pharmacological substance. Let us
make the further assumption that some speaker is deliberately trying to con-
vey this information. This might well be odd in, for instance, the context of a
sermon. Is this linguistic meaning? (In the case described, it is certainly a kind
of meaning, and language is used to convey it.) Presumably it is not linguistic
meaning, but how do we exclude it? One way is to stipulate that linguistic
meaning must either be conventionally associated with the linguistic forms
used, or be inferable from the latter in conjunction with contextual knowledge.
26. Introduction 13
One indication that the above example is not of this type would be its
insensitivitv to the actual words used.
1.6.2 What arewe trying to achieve?
I.6.2.I Specifying/describing meanings
A very important task is to discover a way of specifying or describing mean-
ings, whether of isolated words or sentences, or of utterances in context. The
position taken in this book isthat in general, meanings are not finitely describ-
able, so this task boils down to finding the best way to approximate meanings
as closely as is necessary for current purposes (lexicographers have long had to
confront this problem for words).
I.6.2.2 How meaning varies with context
The meanings of all linguistic expressions vary with the context in whichthey
occur. For instance, the shade of colour indicated by a redhead and red wine
are markedlydifferent; the periods of time denoted by monthin (I) and (2) are
quite likely to be different:
(I) He's here for a month. (could be four weeks; not dependent on time of
utterance)
(2) He's here for the month. (will depend on time of utterance, but could be
31days)
Some variations, like the sex of the doctor in Our doctor hasjust married a
policeman and Our doctor has just married an actress can be predicted
by general principles; other variants are less, or not at all predictable.
Semanticians seek a revealing account of contextual variation.
I.6.2.3 Kinds of meaning
There are different sorts of meaning, each with different properties. For
instance, whatever the difference in meaning between (3) and (4), it does not
affect the truth or falsity of the statement:
(3) Old Joshua Hobblethwaite popped his clogs lastweek.
(4) Old Joshua Hobblethwaite passed away last week.
I.6.2.4 What happens when meanings combine?
Another vital aspect of semantics is how simple(r) meaningscombine to form
more complex meanings. To some extent this is a function of grammatical
structure: for instance, the way redand hat combine in a redhat is not the same
as the way turn and red combine in to turn red. But differences occur even
within the same grammatical construction: the mode of combination of red
and hat in a red hat is different from that of long and eyelash in long eyelashes
(compare long eyelashes and a long river).
27. 14 Meaning in language
I.6.2.5 Systematicity and structure; possibility of formalization
All semanticians are to some extent looking for regularities and system in the
way meanings behave, as this leads to maximally economical descriptions. The
most dedicated to this aspect of semantics are those who attempt to model
the semantic behaviour of natural language expressions by means of a strict
logical or quasi-mathematical formalism. This route will not be followed in
this book.
I.6.2.6 New meanings from old
A striking feature of linguistic expressions is their semantic flexibility: beyond
their normal contextual variability, they can be bent to semantic ends far
removed from their conventional value, witness She swallowed it hook, line and
sinker or You'll find her in the telephone book. The study of such extensions of
meaning is an important task for semantics.
I.6.2.7 Role(s) of context
It is usually assumed that linguistic expressions can be assigned some sort of
context-independent semantic value, although there is much disagreement
regarding exactly what this is. There is also general agreement that context is
of vital importance in arriving at the meaning of an utterance. The role of
context ranges from disambiguating ambiguous expressions as in Wejust got
to the bank in time, through identification of referents (who is he, where is
there, in time for what, in He didn't get there in time), to working out 'between
the lines' messages like B's ignorance of the whereabouts of the corkscrew in:
(5) A: Where's the corkscrew?
B: It's either in the top drawer in the kitchen, or it's fallen behind the
piano.
1.6.3 Theapproachadopted in this book
We are not yet in a position to rule out any approaches which yield insights,
even if some such approaches appear at first sight incompatible. This book
therefore takes an ecumenical position on many issues. In so far as there is a
theoretical bias, it is towards the cognitive semantic position. This means, in
particular, that the meaning of a linguistic expression is taken to arise from the
fact that the latter gives access to a particular conceptual content. This maybe
of indeterminate extent: no distinction is made between linguistic meaning
and encyclopaedic knowledge.
Since this book is not intended to propound a body of theory, but to
acquaint non-specialists with the range of semantic phenomena in language,
there is a bias towards descriptive coverage at the expense of theoretical
rigour.
28. Introduction 15
1.7 Branches of the study of meaning in language
The following are the main broadly distinguishable areas of interest in the
study of meaning. They do not by any means form watertight compartments:
there are many points of overlap.
1.7.1 Lexicalsemantics
Lexical semantics studies the meanings of words; the focus here is on 'content'
words like tiger, daffodil, inconsiderate, and woo, rather than 'form'/
'grammatical' words like the, of, than, and so on. To a non-specialist, the
notion of meaning probably has a stronger link with the idea of the word than
with any other linguistic unit: words are, after all, what are listed in dictionar-
ies, and the main function of a dictionary is to tell us what the listed words
mean. For this reason, lexical semantics perhaps provides the easiest access
route into the mysteries of semantics in general, and this is one reason why it
has been given a prominent place in this book, and whyit comes early.
1.7.2 Grammatical semantics
Grammatical semantics studies aspects of meaning which have direct rele-
vance to syntax. This has many manifestations, which can only be briefly
illustrated here. One problem is the meaning of syntactic categories (problem-
atic, because not everyone believes they can be assigned meanings). Consider,
for instance, the differences in the meaning of yellow in thefollowing:
(6) Shewore a yellow hat. (adjective)
(7) They painted the room a glowingyellow.(noun)
(8) The leavesyellowrapidly once the frosts arrive. (verb)
Another aspect of grammatical semantics is the meaning of grammatical
morphemes like the -ed of walked, the -er of longer, the re- and the -al of
retrial, and so on.
Clearly this overlaps with lexical semantics, partly because some grammat-
ical elements are words (like the, and of), but more particularly because some
aspects of the meanings of full lexical items determine to some degree their
grammatical behaviour (for instance, the fact that / am studying that question
is grammatical, but not I am knowing the answer to that question).
1.7.3 Logical semantics
Logical semantics studies the relations between natural language and formal
logical systems such as the propositional and predicate calculi. Such studies
usually aim at modelling natural language as closely as possible using a tightly
controlled, maximally austere logical formalism. It is arguable that sometimes
such studies shed more light on the formalismused than on the language being
29. 16 Meaning in language
modelled; none the less, valuable insights have come from this approach. To
date, most such studies have concentrated on the propositional/sentential level
of meaning, and have rarely attempted to delve into the meanings of words.
1.7.4 Linguistic pragmatics
For present purposes, pragmatics can be taken to be concerned with aspects of
information (in the widest sense) conveyed through language which (a) are not
encoded by generally accepted convention in the linguistic forms used, but
which (b) none the less arise naturally out of and depend on the meanings
conventionally encoded in the linguistic forms used, taken in conjunction with
the context in which the forms are used. This rather cumbersome formulation
is intended to allow into pragmatics things like the identity of the individual
referred to by John in / sawJohn today, and the assumption that the room in
question had several lights in John entered the room; all the lights were on, at
the same time excluding, for instance, the possibility that the person saying /
saw John today had a private ad hoc arrangement with the hearer that when-
ever he said John, he should be taken to mean "Mary" (since it does not arise
naturally out of the normal meaning of John), and excludingalso the possibil-
ity of someone's inferring from a speaker's slurred speech that they were
drunk (since this does not depend on the conventional meanings of the words
uttered). Pragmatics is usually contrasted with semantics, which therefore
deals with conventionalized meaning; obviously, the three divisions discussed
above belong to semantics.
Suggestions for furtherreading
Much fuller accounts of the semiotic environment of spoken language can be
found in Argyle (1972), Beattie (1983), Ellis and Beattie (1986) and Clark
30. CHAPTER 2
Logical matters
2.1 Introduction
2.2 Argumentsand predicates
2.3 Sense, denotation, and reference:
intension and extension
2.4 Sentence, statement, utterance,
and proposition
2.5 Logical properties ofsentences
19
19
21
22
28
2.6 Logical classes
2.7 Logical relations
2.8 Quantification
2.9 Useand mention
Discussion questions and
exercises
Suggestions for further reading
32
34
35
36
37
39
32. CHAPTER 2
Logicalmatters
2.1 Introduction
This chapter introduces a set of fundamental conceptual tools, mostly drawn
from the field of logic, which, because of their widecurrency in discussions of
semantic matters, constitute indispensable background knowledgefor a study
of meaning in language. The level of treatment here is fairly elementary; some
of the notions introduced will be further refined in subsequent chapters.
2.2 Argumentsandpredicates
A closely linked pair of concepts which are absolutely fundamental to both
logic and semantics are argument and predicate.No attempt will be made here
to explore the philosophical background and underpinning of these notions:
the basic notions are fairly accessible and they willbe treated in an elementary
fashion. Put simply, an argument designates some entity or group of entities,
whereas a predicate attributes some property to the entity denoted by the
argument, or a relation between the entities denoted by the arguments, if there
is more than one. Thus, in John is tall, we can identify John as the argument,
and is tall as the predicate. In John likes Mary, both John and Mary are
arguments, and likes is the predicate which attributes a particular relationship
between the entitiesdenoted by the arguments; in John gave Mary a rose, there
are three arguments, John, Mary and the rose,with gave as the predicate. The
combination of an argument and a predicate forms a proposition: notice that
a proposition may have only one predicate, but may have more than one
argument. It is not clear whether there is any theoretical upper limit to the
number of arguments a predicate may take, but the most one is likely to
encounter in linguistic semantic discussions is four, exemplifiedby Mary paid
John £500for the car:
Arguments: Mary, John, £500,thecar
Predicate: paid (for)
33. 20 Meaning in language
An argument may have a more or less complex internal structure. For
instance, a whole proposition, itself possessing argument(s) and predicate,
may constitute an argument, as in John wassurprised that the man wastall.
Arguments: (i) John (ii) that the man wastall
Predicate: was surprised
Argument (ii) Argument: the man
Predicate: wastall
There are various ways of incorporating propositions as constituents of
complex arguments, and there is no limit to the resulting degree of complexity.
An account of this is beyond the scope of this book.
Predicates are commonly described as one-place, two-place, three-place, etc.
according to the number of arguments they take, so that, for instance, ispoor
is a one-place predicate, and teach is a three-place predicate (John taught Mary
French). But what does it mean to say that teach is a three-place predicate?
How do wedetermine how many places a predicate has?
This is, in fact, a very difficult question, but we can get some handle on it by
looking at a few verbs. Let us start by looking at teach. One aspect of the
problem is immediately obvious when we look at examples such as the
following:
(1) John taught Mary French.
(2) John taught French for two years.
(3) A: What does John do?
B: I think he teaches at Lowhampton High.
(4) Anybody who teaches teenagers should get double salary.
Do we say that teach is three-place in (1), two-place in (2) and (4), and one-
place in (3)? While there is some justification for such an analysis, there is
intuitively a sense in which the (overt) argument structure of (1) is basic, and
irreducible. Logically, for an act of teaching to take place, there must be some-
one who does the teaching, someone who undergoes the teaching (whether or
not they actually learn anything!), and some item of knowledge or skill which
it is hoped will be acquired by the latter. Without at least one each of these
requirements, the notion of teaching is not logically coherent. What, then, are
we to make of (2)-(4) above? It seems that we assume that the missing argu-
ments could in principle be supplied, but the speaker has not supplied them,
presumably because they are not currently relevant, or perhaps in some cases
because they are extremely obvious. In (2), we do not have a vision of John
discoursing in solitude on the French language; nor do we imagine that John
in (3) does something like sneezing, which needs neither audience nor topic.
By the same sort of criteria, read is a basically two-place verb, in spite of the
existence of sentences such as John is reading and John is learning to read. We
cannot make sense of read unless wehave someone to do the reading and some
coded signal to decode. As a final example, consider buy. This requires four
34. Logical matters 21
arguments, as in John sold the car to Mary for £500. If John receives no money,
he is simply giving the car away; if there is no car, Mary is just giving John
some money; if there is no one to receive the money and concede ownership of
the car, then Mary is throwing her money away and taking possession of the
car. And so on. In this way, it is usually possible to determine a basic logically
minimum number of arguments for a predicate.
But our problems are not over. Acts of reading, teaching, and selling (and
sneezing) take place at particular times and particular places. Wemay say that
unlocated, timeless acts of reading, etc., are logically impossible. It is also
necessarily the case that the ambient temperature has such and such a value.
Whereas it is relatively easy to establish the minimum number of arguments for
a predicate, how do we establish a maximum? Or perhaps there is no such
thing? After all, if we learn that John sneezed, we assume that the event
happened at a particular place and time, and that even if the speaker did not
specify these items, they could in principle be specified, along with countless
other things. What is the difference in status, therefore, between the time and
place of John's sneezing, and the subject that John teaches? One approach is
to say that although a verb like sneeze in a sense conjures up a rather complex
picture of a person in a setting acting in a certain way, it highlights only
certain aspects of that scene, but without obliterating or denying the rest.
These highlighted aspects are what distinguish the act denoted from other
possible acts. This means that if we wish to ascertain whether a John-sneeze-
event has occurred, we need only observe happenings pertaining closely to
John; we do not need to check time, place, or temperature. In this way we can
check that the minimawe previously established for the number of arguments
a predicate takes are generally also maxima.
2.3 Sense, denotation, and reference: intension and extension
Language is used to communicate about things, happenings, and states of
affairs in the world, and one way of approaching the study of meaning is to
attempt to correlate expressions in language with aspects of the world. This is
known as the extensional approach to meaning.
The thing or things in the world referred to by a particular expression is its
referent(s): in saying The cat's hungry, I am (normally)referringto a particular
cat, and that cat is the referent of the expression the cat. The whole utterance
attributes a particular state to the cat in question. We can also consider the
whole class of potential referents of the word cat, namely, the class of cats.
This, too, is sometimes called the reference of the word cat. But this is clearly
different from the designation of particular individuals as in the case of The
cat's hungry, so, to avoid confusion, we shall follow Lyons and say that
the class of cats constitutes the denotation of the word cat. So, in the case of
35. 22 Meaning in language
The cat's hungry, the word cat denotes the class of cats, but the cat refers to a
particular cat.
The alternative to an extensional approach to meaning is an intensional
approach. Take the word cat.Whydo weuseit to refer to cats, rather than,say,
to platypuses or aardvarks or spiny anteaters? One answer is that the word is
associated with some kind of mental representation of the type of thing that it
can be used to refer to, and aardvarks do not fit the description associated with
the word cat. This representation constitutes what is called the sense of the
word (or at least part of it). We shall assume in this book that the main
function of linguistic expressions is to mobilize concepts, that concepts are the
main constituents of sense, and that sense (and hence concepts) constrains
(even if it does not completely determine) reference. (It should be noted that
some authors, for instance Lyons, understand sense in a different way. For
them, sense is a matter of the relations between a word and other words in a
language. So, for instance, the sense of cat would be constituted by its relations
with other words such as dog(a cat is necessarily not a dog), animal(a cat is an
animal), miaow (Thecat miaowed is normal but ?The dog miaowed is not).)
2.4 Sentence, statement, utterance and proposition
A number of distinctions need now to be made which at first sight might seem
to be academic nit-picking of the worst sort. However, they are absolutely
vital for clarity in semantics. For the sake of simplicity of exposition, the
distinctions will first of all be explained in connection with declarative sen-
tences only;how the variousnotions apply to non-declarative sentenceswill be
dealt withlater.
2.4.1 Sentence meaning;truth conditions; propositionalcontent
A sentence is a grammatical unit, that is, it is a string of words of a particular
type, whose well-formednessconditions are specified in the grammar of the
language. Thus, The cat sat on the mat and John put his hat on the table are
sentences of English; John put on the table and Is the of mother boy swim are
not. We shall not spend too much time on discussing what distinguishes a
sentence from other grammatical units, such as a phrase, or a word, except to
say that a sentencemust contain at least one independentclause (i.e., one that
does not need another clause, to be grammatical), and a clause must contain at
least one predication (i.e., an argument-predicate structure). The sentence is
thus the smallestlinguisticunit that can be used in an act of 'telling'. Compare
the following for normality/oddness:
(5) A: Tell me something nice.
B: Chocolate.
36. Logical matters 23
A: What do you mean?
B: Well,chocolate is nice.
(6) A: Tellme something nice.
B: Love is a many-splendoured thing.
A: Ah! How true!
The oddness of B's reply in (5) consists in the fact that it neither overtly forms
a sentence, nor permits a plausible sentence to be reconstructed from the
context, as in, for instance, (7):
(7) A: He asked me what I wanted.
B: What did you tell him?
A: Icecream.
Here, the full act of telling is reconstructible as I want ice cream, which
involves a sentence. Henceforward we shall use our intuitions as to what con-
stitutes a sentence.
Notice that we said above that the sentence was 'used in the act of telling':
the sentence itself, on its own, does not in fact tell us anything. What does the
sentence The cat sat on the mat tell us? Is it true? There is no way of knowing,
or rather, the question does not make sense: sentences of themselves do not
necessarily have truth values (some, arguably, have: A molecule of water con-
tains two hydrogen atomsand one oxygen atom).Yet presumably no one would
wish to say that The cat sat on the mat was meaningless. Let us introduce the
notion sentence meaning to designate the semantic properties a sentence pos-
sesses merely by virtue of being a well-formed sentence (of English), before
any question of context or use arises. A sentence possesses this meaning
exclusively by virtue of the words it contains, and their grammatical arrange-
ment. (I assume, here, that words have at least some context-independent
conventionally assigned semantic properties.) We may assume that the gram-
mar of a language is associated with principles of composition, that is, rules
which tell us how to put together the meanings of the constituents of a con-
struction to get the global meaning of the construction. Thus, in The big cat
sat on the smallmat we know, for instance, that smallness is attributed to the
mat and not to the cat, and that a superior vertical position is attributed to the
cat; we know, furthermore, what sort of animal is involved, and that only one
of them would be intended as a referent in any actual use of the sentence. And
so on. In a fairly obvious sense, the meaning of a sentence will constrain the
uses to whichit can be put, at least without the setting up of additional adhoc
conventions. (For example, // / say How are things?, you will know it's my
husband on thephone: that is not what How are things?, by general convention,
means.)
Although a sentence, outside of particular uses, does not have a truth value,
it does have truth conditions, that is, conditions which must hold for the
sentence to be used to make a true statement. Thus, before we can truthfully
37. 24 Meaning in language
say, on some occasion, The cat is on the mat, there must be some relevant feline
occupying a specific position relative to an appropriate item of floor-covering.
Those aspects of the meaning of a sentence which determine whether a state-
ment the sentence is being used to make, in a particular situation, is true or
false, are collectively known as the propositional content of the sentence. Two
sentences with identical propositional content will yield statements with the
same truth values on all occasions of use, as for instance: John caressed Mary
and Mary was caressed by John. By the same token, if two sentences have
different propositional content, there will necessarily exist some conceivable
situation in which they will yield statements with opposite truth values.
Propositional content does not by any means make up the whole of sen-
tence meaning, as conceived here. For instance, the interrogativemeaning of a
question is not included, nor the imperative meaning of a command (more
generally, illocutionary meaning falls outside propositional content). The
force of such words as yet, still and already (sometimes called conventional
implicature) is excluded; the following two sentences, for instance, have the
same propositional content, since their truth conditions are the same, but one
would not want to say that they were identical in meaning (nor are they
appropriate in the same circumstances):
(8) John has not arrived.
(9) John has not yet arrived.
Other aspects of meaning which do not fall under propositional content are
expressive meaning:
(10) It's very cold in here.
(11) It's bloody cold in here.
and features of register such as the formal/colloquial distinction:
(12) My old man kicked the bucket yesterday.
(13) My father passed awayyesterday.
To qualify as part of sentence meaning, a semantic property simply has to
be a stable conventional property of some linguisticexpression. We shall not
even demand that it be manifested on every occasion of use of the expression
in question: it is sufficient that it be potentially present, and not attributable to
context. We shall be particularly impressed by evidence that expressions with
closely similar propositional content do not possess the property in question.
An example of this is the potential expressive nature of the word baby
compared with infant (see Chapter 3 for a more extensive discussion of
non-propositional meaning).
2.4.2 Statement meaning
One does not make a statement simply by producing a sentence of declarative
38. Logical matters 25
form. Someone in a language class, for instance, practising the tenses of Eng-
lish, who says:
Thecat sat on the mat.
The cat sits on the mat.
The cat will sit on the mat.
is not making a series of statements. The first requirement for a statement to
have been made is that a proposition must have been expressed; the second
requirement is that an appropriate commitment be made to the truth of the
proposition. Let us dwell on this for a moment, beginningwith the notion of a
proposition.
2.4.2.1 Propositions
A simple proposition attributes some property to an entity, or a relation
between two or more entities. It is either true or false (even if it is not practic-
ally or even physically possible to ascertain which):truth or falsehood is a sign
that at least one proposition has been expressed. A proposition is not a specif-
ically linguisticentity (although we shall not dwelltoo long on the problem of
what sort of entity it is, and in what Platonic or other realm it subsists). The
same proposition may be expressed by an indefinitely large number of
sentences:
John sawMary.
John sawhis sister.
Mary was seen by Peter's uncle.
etc.
These can all express the same proposition, provided, of course, that Mary is
John's sister, and that John is Peter's uncle, and so on.
Why, then, is The cat sat on the mat not a proposition? As it stands, it is
neither true nor false. It becomes true or false when it is asserted of some
specific cat and some specific mat. Until definite referring expressions in a
sentence have been assigned referents, it does not express a specific prop-
osition. A sentence like The cat sat on the mat can be used to express an
indefinitely large number of different propositions (i.e., with reference to dif-
ferent cats and different mats) on different occasions of use.
2.4.2.2 Epistemic commitment
A proposition may be 'entertained', without any stance being adopted
towards its truth or falsehood, as, for instance, in a logic class, where proposi-
tions are entertained, and their interrelationships studied, such as
(14) All aardvarks are purple.
Tinkerbelle is an aardvark.
Hence, Tinkerbelle is purple.
39. 26 Meaning in language
However, a proposition on its owncannot actually communicate anything: it is
not an item of knowledge. To communicate, it must be energized with some
kind of illocutionary force. For instance, the proposition expressed by The
earth isapolyhedron is of no interest until someone claims it is true (or false): a
true or false proposition is an item of knowledge. The illocutionary force in
such a case would be that of assertion. A (minimum) statement is therefore a
proposition uttered, as Lyons puts it, 'with epistemic commitment'. There are
other possibilities for illocutionary force, to be studied in due course.
2.4.2.3 Incomplete propositions
Consider the following exchange:
(15) A: What's the time?
B: Half-past four.
Is B telling A anything? Obviously he is. Then what proposition is he express-
ing? Again the answer is obvious: "The time is 4.30". But this is not what he
actually says. In such cases, it is up to the hearer to reconstruct the full form of
the proposition on the basis of contextual clues, but there is no doubt that it is
the full form of the proposition that the speaker intends to convey, and this
should form part of statement meaning.
We are now in a position to spell out what statement meaning consists of.
First of all, statement meaning incorporates in its entirety all aspects of sen-
tence meaning which belong to the sentence used in making the statement.
Secondly, statement meaning includes a specific proposition or propositions,
the identification of which requires that (a) referents be assigned to any def-
inite referring expressions in the sentence, and (b) incomplete propositions
be completed. It is assumed that the proposition(s) is expressed with the
appropriate epistemic commitment.
2.4.3 Utterance meaning
Even a fully elaborated statement meaning may not adequately represent
everything the speaker intended to convey by uttering the sentence to which it
applies. To take a verysimple example, consider the following:
(16) A: Haveyou cleared the table and washed the dishes?
B: I've cleared the table.
In normal circumstances, it would clearly be part of B's intended message that
s/he had not washed the dishes. Yet this cannot be obtained by elaborating or
completing the proposition expressed. This is even clearer in the following case
where it is obviously the speaker's intention to convey the proposition that A
is too late for supper:
(17) A: Am I in time for supper?
B: I've cleared the table.
40. Logical matters 27
These extra (i.e. covert) propositions are expected to be inferred by the hearer
on the basis of contextual information, but they go well beyond the mere
filling out of missing bits in what is actually said. We shall give the name
utterance meaning to the totality of what the speaker intends to convey by
making an utterance, within certain necessary limits.
Two people might, for instance, have an arrangement such that How was
your day? means "How was your day?", but How was the day today? means
"My husband is going to his club this evening—we'llhave plenty of time for
fun". Wewould wish to exclude this sort of meaning from utterance meaning.
I think the key point is that such a use requires a special ad hoc stipulation
on the part of speaker and hearer: the hearer's understanding of the clue does
not arise entirely from his general knowledge of the rules and conventions
governing the use of the forms in question.
Once again, utterance meaning subsumes statement meaning, but only in
the sense that the latter must be traversed in order to arrive at the former, i.e. is
a necessary step in derivation: utterance meaning does not necessarily
incorporate statement meaning as a proper part (although, of course, it may
do).
2.4.4 Non-declaratives
So far, we have looked only at declarative sentences. However, the notions
elaborated above apply equally to non-declarative sentences. Take the case of
a question. The general notion of sentence meaning is probably unproblematic
here. The equivalent to statement meaning will be question meaning. Recall
that the propositional content of a statement determined whether it was true
or false in a given situation. Now questions do not have truth values, but they
do have propositional content. Although a question does not have a truth
value, a given question-statement pair has what might be called an answer
value, that is, the statement is or is not an answer to the question, and if it is,
it is either true or false. This answer value is determined equally by the
propositional content of the question and the statement:
(18) A: What day is it today? (said on Tuesday)
B: It's Monday today, (false answer)
A: What day is it today? (said on Monday)
B: It's Monday today. (true answer)
A: Who do you lovebest?
B: It's Monday today. (no answer)
Just as a declarative sentence has truth conditions, we can postulate that a
question-and-answer pair, considered as sentences, has answer conditions, that
is,the conditions whichmust hold for the statement to constitute a true answer
to the question.
The same trick can be played with imperatives, except that a linguistic
41. 28 Meaning in language
response is not always, perhaps not usually, what an imperative sets out to
elicit. Generally what is required is an action (in the broadest sense, which
covers such cases as Don't move!). In the case of a fully contextualized com-
mand, where a full command meaningis operative, a given action either counts
as compliance or does not, that is, it has a compliance value, equivalent (in the
relevant respects) to the truth value of a statement. What its compliance value
is, is determined by the propositional content of the command. An imperative
sentence has a set of compliance conditions, but no compliance value (how do
you obey Put it there,withoutknowingwhat it is, or where thereis?).
Let us now try to summarize and generalize. Declarative sentences, in gen-
eral, do not, in themselves, make statements, interrogative sentences do not
ask questions, and imperative sentences do not issue commands. These only
result when the sentences are fully contextualized. Each of these, however, has
propositional content, which governs what illocutions it can be used to per-
form when properly contextualized, that is, which statements can be made,
questions asked, and commands issued, etc.
2.5 Logicalpropertiesof sentences
2.5.1 Logicalrelations betweensentences
A number of logical relations between sentences make a not infrequent
appearance in semantic discussions, especially in lexical semantics. They are
related to, but not identical to, relations that logicians recognize between pro-
positions. It is, however, necessary to bear in mind their logical basis, if confu-
sion is to be avoided. Five relations will be recognized here: implication/
entailment, equivalence,contrariety, contradiction, and independence.
2.5.1.1 Entailment
This and the following relations strictly speaking hold between propositions,
and that is how we shall treat them to begin with; the connection with sen-
tences will be dealt with later. Entailment is the relation which holds between
the P and the corresponding Q items in the following:
P
It's a dog.
John killed the wasp.
All dogs are purple.
Q
It's an animal.
The wasp died.
My dog is purple.
Notice that we are assuming that these sentences express propositions; that is
to say, the sentences are being used in a particular context with particular
reference. We are further assuming that in the first pair it refers to the same
entity in each case, and in the second pair, the same wasp and the same event
are being referred to. To say that proposition P entails proposition Q means
42. Logical matters 29
that the truth of Q follows logically and inescapably from the truth of P, and
the falsity of P follows likewise from the falsity of Q. So, in the first pair of
sentences above, if it is true of some entity that it is a dog, then it follows
ineluctably that it is an animal, and if it is not an animal, then there is no way
it can be a dog. Similarly,in the second pair, if John killed some wasp, then we
cannot avoid the conclusion that the wasp died, and if the wasp did not die,
then it cannot be the case that John killed it.
Entailment, as used by linguistic semanticists, is to be distinguished from
what logicians call material implication. A proposition P materially implies
another proposition Q if and only if (henceforward iff) it is never the case that
P is true and Q false. At first sight this seems to be essentially the same as
entailment. However, there is a crucial difference: the definition of material
implication makes no reference to the meanings of the propositions, merely to
a relation between their truth values; entailment, on the other hand, (some-
times called strict implication) is essentially a relation between meanings. To
illustrate this point, consider the propositions It's a dog and All bachelors are
unmarried. It can never be the case that the first is true while the second is false,
so wehave a case of material implication. But this is not entailment, because it
is not a consequence of the meaning relations between the two propositions,
but of the fact that the second proposition cannot under any circumstances
(except by altering the meanings of the components) be false. On the other
hand, although it is true that It's a dog materially implies It's an animal,
because the former cannot be true while the latter is false, it is also the case
that the former entails the latter, because the truth-value relation holds by
virtue of the relation in meaning between dog and animal.Material implica-
tion is essentially of no interest to linguistic semantics, although the status of
All bachelors are unmarried as being always true is of interest (see 2.5.2.1
below: analytic expressions).
Strictly speaking, entailment does not hold between sentences, because
sentences do not have truth values. However, one frequently encounters
in semantic texts statements to the effect that such-and-such a sentenceentails
some other sentence. This can be taken as a kind of shorthand for something
slightly more complex. Saying that sentence S1
entails sentence S2
means that
in any context where S1
expresses a true proposition, S2
also necessarily
expresses a true proposition, provided that corresponding definite referring
expressions in the two sentences are co-referential. Obviously, in the case of
It's a dog and It's an animal, the two occurrences of it must refer to the same
entity for the logical relation to hold, and in the case of John killed the wasp
and The wasp died, we must be talking about the same wasp, and the time
references must be the same.
Two other properties of entailment must be emphasized. The first is that the
relation is not determined by context: it is context independent, since it
depends entirely on the meanings of the constituents of the sentences. Con-
sider a case where John has in front of him a box of coloured disks, in which
43. 30 Meaningin language
all the red disks are round in shape, and all the green disks are square. In such
circumstances, the truth of John picked a square disk from the box follows
inescapably from the truth of John pickeda green diskfrom the box.But clearly
this relation of truth values does not arise from relations between green and
square, but from the context: it would in principle have been just as easy to
have all the red disks square, and the green disks round. On the other hand, the
relation between It's a dog and It's an animalis independent of any particular
contexts.
The second property is that the truth of the entailed sentence must follow
inescapably from the truth of the entailing sentence. It is not enough for it to
be usually true, or even almost always true; it has to be unthinkable that it
might be false. Consider the relation between It's a dog and (i)It'sapet and(ii)
It canbark. Most dogs that most people encounter are pets, but there are such
things as wild dogs, so the relationship is merely one of expectation. This is not
entailment. Likewise in the case of (ii),most dogs can bark, but a dog with a
defective larynx does not thereby cease to be a dog,so the relationship is not
logically necessary. Only logically necessary, context-independent relation-
ships count as entailment. (Weshall modify this position in Chapter 3, but for
the moment it stands.)
2.5.1.2 Equivalence
Propositional equivalence between two sentences can be straightforwardly
defined as mutual entailment. That is, in effect, equivalent to saying that the
two sentences always express the same proposition (provided, of course, that
corresponding definite referring expressions are co-referential). The following
are examplesof equivalence:
John killedthe wasp.
The wasp is dead.
It began at 10o'clock.
The wasp was killedby John.
The wasp is not alive.
It commenced at 10o'clock.
If it is true that John killed the wasp, then it is also true that the wasp was
killed by John and if it is true that the wasp was killed by John, then it is also
necessarily true that John killed the wasp; a parallel two-wayentailment holds
between the members of the other two pairs.
2.5.1.3 Contrariety
Contrary propositions may not be simultaneously true, although they may be
simultaneously false. The followingare examples:
John killedthe wasp.
John killed the wasp.
This paint is red.
The waspisalive.
Mary killedthe wasp.
This paint is green.
It cannot be simultaneously true of some wasp both that John killed it and
that it is still alive; on the other hand, if the wasp is actually dead, but it was
44. Logical matters 31
Bill who killed it, then both John killed the wasp and The wasp is alive are false.
We can define this relation in terms of entailment, by saying that S1
and S2
are
contraries iff S1
entails not-S2
, but not-S2
does not entail S1
(and vice versa).
Thus, This paint is redentails This paint is not green, but This paint is not green
does not entail Thispaint is red,since it might well be, for instance, yellow.
2.5.1.4 Contradiction
Contradictory propositions must have opposite truth values in every circum-
stance: that is, they cannot be either both true or both false. In any particular
circumstance, one member of a contradictory pair must be true and the other
false. The following sentencesexemplify contradictory pairs:
The wasp is dead.
John is still singing.
No dogs are brown.
The wasp isalive.
John isno longer singing.
At least some dogs are brown.
If John is still singing, then it is false that he is no longer singing; to this extent,
this is like contrariety. However, there is a crucial difference: if it is false that
John is still singing, then it must be the case that he is no longer singing, and if
it is false that he is no longer singing, then he must be still singing. We can
define contradiction in terms of entailment, by saying that S1
and S2
are
contradictories iffS1
entails not-S2
, and not-S2
entails S1
(and viceversa).
2
.
5
.
1
.
5 Independence
For some pairs of propositions, the truth values vary independently of one
another: they maybe both true, both false, or one true and the other false:
John isretired.
It isTuesday today.
Mary is married.
Christmas day falls on a Wednesdaythisyear.
The relations described in this section havean important role in the analysis of
meaning relations between words, as weshall seein later chapters.
2.5.2 Analytic, paradoxical, and synthetic sentences
2.5.2.1 Analyticity
Analytic sentences are sentences which automatically express true proposi-
tions in any context, by virtue of the meanings of their constituent words and
their arrangement. The following sentences are therefore analytical:
Bachelors are unmarried.
John's uncle is a man.
This proposition iseither true or false.
2.5.2.2 Paradox
Paradoxical sentences automatically express false propositions:
45. 32 Meaning in language
Bachelors are married.
John's sister isa man.
This red paint isgreen.
2.5.2.3 Syntheticity
Synthetic sentences are those which express true propositions in some (con-
ceivable) contexts (although they may be false of the world as weknowit) and
false ones in others (this is the normal kind of sentence used in
communication):
John's sister is married.
This paint isgreen.
All dogs are brown.
(The last sentence is actually false, but it is not logically false; it is easy to
imagine circumstances in whichit would be true.)
2.6 Logical classes
2.6.1 Class relations
2.6.1.1 Identity
Two classes C1
and C2
are said to be identical if everything that belongs to C1
also belongs to C2
, and vice versa. Thus, the class of fathers and the class of
male parents are identical, as are the class of pairs of spectacles and the class
of pairs of glasses (on the relevant interpretation of glasses).
2.6.1.2 Inclusion
Class C1
is said to include class C2
if everything that isa memberof C2
is also a
member of C1
, but not vice versa. Thus, for instance, the class of animals
includes the class of dogs, the class of aardvarks, etc. The set of dogs is
described as a subclass of the set of animals, and the set of animals as
a superclass of the set of dogs. (Inclusion is defined here so as to exclude
identity; it can be defined so as to include identity.)
2.6.1.3 Disjunction
Classes C1
and C2
are said to be disjunct if no member of C1
is also a member
of C2
. The class of cats and the class of aardvarks are disjunct in this sense, as
are the class of red things and the class of green things.
2.6.1.4 Intersection
Classes C1
and C2
are said to intersect if they have some membersin common,
but each has members which do not belong to the other (i.e. complete inter-
46. Logical matters 33
section, or identity, is excluded here). The class of red things and the class of
round things intersect in this fashion, as do the class of architects and the class
of amateur musicians. The set of common members to two (or more) overlap-
ping classesis often referred to as the intersection of the two (or more) classes.
2.6.1.5 Union
The combined set of members belonging to either of two (or more) classes
(including overlapping classes) is called the union of the two (or more) classes.
Thus the union of the class of dogs and the class of cats is constituted by the
class of all entitieswhich are either cats or dogs.
2.6.2 Class relationsand propositionalrelations
There are obvious connections between the class relations described above and
the propositional relations described earlier. For instance, the fact that It's a
dog entails It's an animal is not unconnected to the fact that the class of dogs is
a subclass of the class of animals (although the connection is not a necessary
one unlesswedefine the class of dogs as "the class of all possible dogs"). Some
relations emerge more naturally within one approach than another. For
instance, we gave no propositional parallel for class overlap. This could be
called propositional independence, since the truth of either proposition in
such a pair imposes no logical restraint on the truth value of the other, as in
the case of John is an architect and John is an amateur musician.The relation
of contradiction, on the other hand, emerges more naturally within the prop-
ositional approach. To find an equivalent of the relation between, say dead
and alive within the class approach, wewould have to say that the class of dead
things and the class of alive things were (i) disjunct and (ii) exhaustive of a
superclass of animate things.
2.6.3 Mapping
It sometimes happens that the members of one class have a relation of cor-
respondence of some kind with one or more members of a parallel class. This
type of correspondence is known as mapping. An examplewillmake this clear.
It is a well-knownfact that a person's fingerprints are uniquely distinctive.If,
therefore, we think of the class of persons and the class of fingerprints, there is
a straightforward mapping relation between the two classes, in that each
member of one class corresponds to a specific member of the other set. This is
known as one-to-one mapping. Contrast this situation with the two classes
FATHERS and CHILDREN. Everymember of the FATHERS class corresponds to one
or more members of the CHILDREN class, but every member of the CHILDREN
class corresponds to a single specific member of the FATHERS class. Here we
have one-to-many mapping between fathers and children, but many-to-one
mapping between children and fathers. Yet another elementary mapping rela-
tion holds between the class of word forms and the class of meanings. If we
47. 34 Meaning in language
allow the possibility of synonymy,then some word forms (e.g. perhaps begin
and commence) will map on to the same meaning, whereas other word forms
(e.g. bank) will map on to more than one meaning. This is known as many-to-
many mapping (see section 7.1.2 for a more detailed consideration of mapping
between words and meanings).
2.7 Logicalrelations
Another useful set of concepts borrowed from logic are to do with relations
between individual entities. The entities may be anything at all: objects, people,
places, ideas. So can the relations be anything: "brother of, "smaller than",
"has played string quartets with", "logically depends on". The logical proper-
ties of such relations can be grouped under four headings: transitivity,
symmetry, reflexivity, and converseness.
2.7.1 Transitivity
A relation that is transitive is one such that if A is related in this specific way to
B and B to C, then it follows inescapably that A stands in the relation to C.
Suppose A, B, and C are people, and the relation is "is taller than". Then if A
is taller than B and B is taller than C, then A is necessarily taller than C. If a
relation is intransitive, then if A stands in the relation to B and Bto C, then it is
logically impossible for A to stand in the relation to C. This is the case with "is
the mother of": if A is the mother of B and B the mother of C, then A cannot
be the mother of C. A relation may be neither transitive nor intransitive;we
shall call such a relation non-transitive. If John has played duets with Bill, and
Bill has played duets with Tom, then we are not in a position to conclude
anything regarding John's musical relations with Tom.
2.7.2 Symmetry
A symmetricrelation is one such that if A stands in a particular relation to B,
then B necessarily stands in that same relation to A. For instance, if A is near
to B, then B is near to A. If a relation is asymmetric, then if A stands in the
relation to B, B cannot stand in the same relation to A. An example of an
asymmetric relation is "is taller than". Once again, it is useful to have a desig-
nation for relations that are neither symmetric nor asymmetric, such as "is
sexually attracted by"; we shall label these non-symmetric.
2.7.3 Reflexivity
The property of reflexivity is not of great usefulness in semantic analysis: it is
included for the sake of completeness. A relation is reflexive if something
necessarily stands in that relation to itself. This includes most types of identity
48. Logical matters 35
relation such as "has the same name as", "is the same age as", etc. Again,we
can recognize irreflexive relations like "is taller than", and non-reflexive
relations such as "knows theweight of.
2.7.4 Converseness
Converseness is a relation between relations. Tworelations are conversesif one
yields the same proposition as the other when the arguments are reversed. By
this criterion, "above" and "below" are converses (perhaps more strictly "is
above" and "is below"), because A isbelow B, assuming constancy of A and B,
expresses the same proposition as B is above A. Other examples of (sentences
expressing) converse relations are: A is B's offspring/B is A's parent, A saw BIB
was seen by A, A sold B to C/C bought B from A (it is usual to disregard
'automatic' adjustments in grammatical realization, such as the change from
to to from in the case of buy and sell). We have defined Converseness in terms
of two-wayentailment between two sentences (e.g. A is taller than B and B is
shorter than A). It is useful to havea relation defined on a one-way entailment.
For instance, A is B's doctor entails B is A's patient, but the reverse entailment
does not hold because other practitioners in the medical and para-medical
field, such as dentists and speech therapists, also have patients. We shall say
that doctor is a semi-converse of patient.
2.8 Quantification
2.8.1 Quantifiers
In standard first-order predicate calculus, propositional functions are con-
strained by quantifiers: these in effect limit the applicability of the predicate to
the argument(s). Classical logic only has two quantifiers, the existential quan-
tifierand the universal quantifier. The existential quantifier says somethinglike
this:
There exists at least one 'x' such that 'x sneezed'.
This is typicallyexpressed in logical notation as:
3x (sneezed (x)).
This could be roughly translated as Someonesneezed. A man sneezed would go
into this speciallogical languageas:
There existsat least one individualx such that x is a man and x sneezed.
3x (sneezed (x) & man (x)).
The universal quantifier corresponds roughly to the ordinary language all,
every. Thus Dogs areanimals would translate as:
49. 36 Meaning in language
For all x, x isa dogentails x isananimal.
Vx (dog (x) —>animal(x) ).
2.8.2 Scope
In the sentence Mary ruffled John's hair and kissed him again we do not know
without further contextual evidence, whether it was only Mary's kissing of
John that was repeated, or the double action of ruffling the hair and kissing.
This is an ambiguity of scope: we do not know how much of the previous
sentence is included in the range of applicability of again. The term scope is
usually used in connection with quantifiers: again is a kind of quantifier over
events. An example involving a more traditional quantifier is: Some womenand
foreigners must register with the police. Here we are uncertain whether only
some foreigners should register, or whether they all should (which would be
the most natural interpretation of Foreigners must register). Reversing the
order of constituents would remove the ambiguity: Foreigners and some
women must register.The possibilities for variations of scope of this sort are
tightly constrained by syntactic structure. (For a more detailed discussion of
quantification, see Chapter 14,section 14.6.)
2.9 Useand mention
Consider the difference between (19) and (20):
(19) Snow has four letters.
Snow is a noun.
Snow is a natural kind term.
Snow is an English word.
Snow iseasy to pronounce.
(20) Snow iswhite.
Snow damages crops.
Snow is frozen water.
The difference between these two sets is usually designated as a difference
between use (here, of the word snow), as in (20), and mention (of the word
snow), as in (19). In the sentences in (19) we are using the word form snow to
identify a word of the language, and we then proceed to say something about
that word; in the sentences in (20) we are using the word form to identify a
substance in the world, prior to predicating something of it. A simple way of
distinguishing the two is to apostrophize (or italicize) the language unit in
question. If this makes a negligible effect on the meaning, then it is a case of
mention:
Snow has four letters.
Snow is an English word.
50. Logical matters 37
*Snow is white.
*Snow damages crops.
Mention may involve any stretch of language:
Go to home is ungrammatical.
*Go to home is ungrammatical.
This brings us to the end of our brief survey of useful logical notions, and
provides an elementary toolkit which will be drawn upon, and sometimes
further refined, as and when the occasion demands.
Discussion questions and exercises
1. Argumentsand predicates
Mark the following predicates as one-, two-, three-, or four-place (1,2,3,4) (think
in terms of semantics rather than syntax):
yawn steal thank pay be tall be taller than meet put Imagine
day-dream cost understand explain
2. Sentence,statement, utterance, and proposition
Of which of the above can the following be said?
X was inaudible.
X was uninformative.
X was false.
X was in a foreign accent.
X was ungrammatical.
X was insincere.
3. For each of the following pairs of sentences, say whether the
propositional contentof the membersis the same or different:
(i) (a) Take your hands off me! (said by a woman to a man)
(b) Take your filthy paws off me! (ditto)
(ii) (a) I always get my bread from Gregg's, because it's cheaper.
(b) I always buymybread from Gregg's, because it's cheaper.
(iii) (a) Don't you find him rather skinny?
(b) Don't you find him rather thin?
(iv) (a) Haveyou read the stuff he wrote about telepathy?
(b) Haveyou read the garbage he wrote about telepathy?
51. 38 Meaning in language
(v) (a) Shewas there at the start of the race.
(b) Shewas there at the beginning of the race.
(vi) (a) John hasn't turned up.
(b) John hasn't turned up yet.
(vii) (a) Old Joshua Hobblethwaite died last week.
(b) Old Joshua Hobblethwaite passed away last week.
4. In which of the following does the (a)-sentence entail the (b)-
sentence? Are there any problems?
(i) (a) X is acat.
(b) X has four legs.
(ii) (a) Xis acat.
(b) X is an animal.
(iii) (a) X is acat.
(b) X is a quadruped.
(iv) (a) X is a quadruped.
(b) X has four legs.
(v) (a) X is a quadruped.
(b) X is an animal.
(vi) (a) X is a pet.
(b) X is an animal.
(vii) (a) X is a pet.
(b) Xis alive.
(viii) (a) Xis not dead.
(b) X is alive.
(ix) (a) X has stopped smoking.
(b) X doesn't smokeany more.
(x) (a) Xtaught YZ.
(b) Y learnt Z.
(xi) (a) X killedY.
(b) Yis not alive.
(xii) (a) X watchedY.
(b) Ywas doing something.
5. Mark the propositional relationship between the members of the
following pairs of sentences as either EQUIVALENCE, CONTRARIETY,
CONTRADICTION, Or CONVERSENESS:
(i) (a) Proposition Pis true.
(b) PropositionPis false.
(ii) (a) John likes Mary.
(b) John dislikes Mary.
(iii) (a) Maryagrees with the statement.
(b) Mary disagrees with the statement.
(iv) (a) Mary borrowedthe book from John.
(b) John lent the bookto Mary.
(v) (a) John killed the wasp.
(b) Thewasp is still alive.
(vi) (a) John is not married.
(b) John is abachelor.
52. Logical matters 39
6. Classify the following relations with regard to their TRANSITIVITY (i.e.,
as TRANSITIVE, INTRANSITIVE,Or NON-TRANSITIVE) and their SYMMETRY
(i.e., as SYMMETRIC, ASYMMETRIC, Or NON-SYMMETRIC):
parent of ancestor of brother of related to sibling of friend of near to
to the right of far from resembles
Suggestions for further reading
The treatment here has been very informal. A similar elementary treatment,
but with more practical exercises, will be found in Hurford and Heasley
(1983). Lyons (1995) develops the philosophical background more fully, but
still at an elementary level. Those requiring initiation into logical formaliza-
tion will find an accessible introduction in Allwood, Anderson, and Dahl,
(1977). Lyons (1977) gives a more detailed treatment of many of the topics
touched on here. Cann (1993), McCawley (1981), and Larson and Segal (1995)
are only for those who are really serious about the application of logic to
language.
54. CHAPTER 3
Types and dimensions of meaning
3.1 Introduction
3.2 Descriptive and non-descriptive
meaning
3.3 Dimensionsof non-descriptive
meaning
43
46
48
3.4 Non-descriptivedimensions
Discussion questions andexercises
Suggestions for further reading
58
62
63
56. CHAPTER 3
Types and dimensions of meaning
3.1 Introduction
The purpose of the present chapter is to survey (albeit somewhat superficially)
the range of possible varieties of meaning in language. Before we can do this,
we need some idea of what is to count as meaning. There are many different
opinions on this question, but the matter will not be argued in detail here,
since many of the divergent views are simply a question of terminology—one
is to some extent at least free to stipulate what is to count. In this book a broad
characterization of meaning will be adopted: meaning is anything that affects
the relative normality of grammatical expressions. This is an example of a
contextual approach to meaning, because relative normality is a concept which
applies only to combinations of elements; that is to say,it implies that meaning
is to be studied by observing the interactions between elements and other
elements, in larger constructions such as sentences. It follows from this
characterization that if two expressions differ in meaning, then this will show
up in the fact that a context can be found in which they differ in normality;
conversely, two expressions with the same meaning will have the same
normality in all contexts. So, for instance, we know that dog and cat differ in
meaning (to take a crudely obvious case) because (for example) Our cat has
had kittens is more normal than ?Our dog has just had kittens. Likewise, we
know that pullover and sweater are at least very close in meaning, because of
the difficulty in finding contexts in which they differ in normality (for further
discussion of synonymy, seeChapter 8). (Note that 'mention' contexts, such as
Pulloverl?Sweater has eight letters, do not count.) It also follows from the
characterization adopted here that the normality profile of a linguistic item,
that is to say, its pattern of normality and abnormality across the full range of
possible contexts, gives in some sense a picture of its meaning. It does not,
however, tell us what meaning really is. This is a deep and controversial ques-
tion; it will be generally assumed in this book that meaning is in essence
conceptual (see Chapter 6), but is most easily studied through language.
57. 44 Meaning in language
3.1.1 Semantic anomaly versus grammatical anomaly
For the characterization of meaning given above to work, we need to be able
to separate semantic anomaly from grammatical anomaly. This is another
contentious issue, but I believeit ispossible to get some grip on it. The account
given here largely follows that given in Cruse (1986).
The most commonly encountered criterion for separating the two types of
anomaly is corrigibility:it is claimed that grammatical anomalies are typically
corrigible in the sense that it is obvious what the 'correct' version should be,
whereas semantic anomalies are typically not corrigible. Thus, *Me seed two
mouses can easily be corrected to / saw twomice,whereas there is no obvious
way of amending *The noiseless typewriter-blasts squirmed faithfully. How-
ever, whilethis may be generally true, it is not difficult to find easily correctable
anomalies which intuitively are clearly semantic:* This hole is too large for
John to crawl through.
There is a basic drawback with the notion of corrigibility, which is that it is
presupposed that one knows what was originally intended. A better approach
is to ask what is the minimum change to the sentence (or whatever) that will
remove the anomaly. There are three possibilities (assuming that the anomaly
has a single source):
(i) The anomaly can only be cured by replacing one (or more) of the full
lexical elements (i.e. a noun, verb, adjective, or adverb). In this case we
can be reasonably certain that weare dealing with a semantic anomaly:
(I) John istoo *small to get through this hole.
Vbig
(ii) The anomaly can only be cured by changing one or more grammatical
elements (affixes, particles, determiners, etc.), but not by changing a
full lexical item. In this case we can be sure that the anomaly is
grammatical:
(2) Mary *begoing home.
Vis
(iii) The anomaly can be cured either by grammatical or by lexical adjust-
ment. In this case we need to know whether the lexical possibilities
form a natural semantic class or not: if they do, the anomaly can be
taken as semantic. Compare (3)and (4):
(3) *Mary went home tomorrow./Mary will go home tomorrow.
(grammatical adjustment)
Mary wenthome *tomorrow.
yesterday.
last week.
etc.
(lexical adjustment)
58. Types and dimensions of meaning 45
There is one more possible diagnostic criterion: a semantic anomaly can
often be improved by manipulating the context, whereas this is usually not
possible with pure syntactic anomalies:
(5) The chair sawMary.
(Mary has a persecution mania. She believes all her accidents are due to
malevolent forces. No doubt the chair saw her, computed her path across
the room, and placed itselfjust whereshewould trip over it.)
No amount of contextual elaboration can reduce the anomaly of The mans
possess threecar.
3.1.2 Typesof anomaly
We have so far treated anomaly as a unitary phenomenon, without trying to
distinguish different sorts. It is quite a useful analytical tool, even without
further refinement, as most speakers have sensitive intuitions regarding the
normality or oddness of a bit of language. But it is sometimes useful to make a
distinction between different types of anomaly. The following are the main
varieties (they are only illustrated here: more detailed discussion will be found
in Chapter 12).
3.1.2.1 Pleonasm
John chewedit with his teeth.
It was stolenillegally.
Mary deliberately made a speech.
These examples give a feeling of redundancy: how else can you chew some-
thing, if not with your teeth? How can anybody make a speech accidentally?
We shall look further into the reasons for pleonasm in a later chapter: for the
moment an intuitivegrasp is sufficient.
In this case the items which remove the anomaly have nothing in com-
mon semantically, and the anomaly of (4) can hence be diagnosed as
grammatical.
(4) *Le livreest sur le table./Le livre est sur la table.
(grammatical adjustment)
Le livre est sur le *table.
fauteuil.
plancher.
buffet.
rocher.
frigo.
Here the items which remove the anomaly share a component of mean-
ing, namely, an indication ofpast time.
